Cell-free Synthesis of Deoxyhypusine SEPARATION OF PROTEIN SUBSTRATE AND ENZYME AND IDENTIFICATION OF 1,3-DIAMINOPROPANE AS A PRODUCT OF SPERMIDINE CLEAVAGE*

The post-translational formation of hypusine (jV-(4-amino-2-hydroxybuty1)lysine) occurs in a precursor of eukaryotic initiation factor 4D by way of two major steps: 1) transfer of the 4-aminobutyl moiety from spermidine to the t-amino group of a specific lysine residue to form an intermediate, deoxyhypusine; 2) hydroxylation of the deoxyhypusine residue to form hypusine. The initial step of this modification, deoxy- hypusine synthesis, was studied in fractionated lysates of Chinese hamster ovary cells, untreated, or treated with a-difluoromethylornithine (DFMO); the enzyme(s) and the protein substrate (eukaryotic initia- tion factor 4D precursor) were separated. The enzyme activity was found in the 0-45% ammonium sulfate fraction from both untreated and DFMO-treated cells. The protein substrate was detected in the 45-7570 ammonium sulfate fraction from cells depleted of spermidine by treatment with DFMO, but not in any frac- tion from untreated cells. Upon further purification of the protein substrate by ion exchange chromatogra- phy, the requirement for a pyridine nucleotide, notably NAD+, became apparent. Free 1,3-diaminopropane was identified as a spermidine cleavage product formed concurrently with the 4-aminobutyl transfer step of deoxyhypusine synthesis. Compounds structur-

The post-translational formation of hypusine (jV-(4amino-2-hydroxybuty1)lysine) occurs in a precursor of eukaryotic initiation factor 4D by way of two major steps: 1) transfer of the 4-aminobutyl moiety from spermidine to the t-amino group of a specific lysine residue to form an intermediate, deoxyhypusine; 2) hydroxylation of the deoxyhypusine residue to form hypusine. The initial step of this modification, deoxyhypusine synthesis, was studied in fractionated lysates of Chinese hamster ovary cells, untreated, or treated with a-difluoromethylornithine (DFMO); the enzyme(s) and the protein substrate (eukaryotic initiation factor 4D precursor) were separated. The enzyme activity was found in the 0-45% ammonium sulfate fraction from both untreated and DFMO-treated cells. The protein substrate was detected in the 45-7570 ammonium sulfate fraction from cells depleted of spermidine by treatment with DFMO, but not in any fraction from untreated cells. Upon further purification of the protein substrate by ion exchange chromatography, the requirement for a pyridine nucleotide, notably NAD+, became apparent. Free 1,3-diaminopropane was identified as a spermidine cleavage product formed concurrently with the 4-aminobutyl transfer step of deoxyhypusine synthesis. Compounds structurally related to spermidine, e.g. caldine, N"-benzylspermidine, homospermidine, and a spermine homologue, thermine, as well as 1,7-diaminoheptane, 1,8-diaminooctane, and 1,9-diaminononane caused significant inhibition of deoxyhypusine synthesis presumably due to competition with spermidine. 1,3-Diaminopropane exhibited a potent inhibition of deoxyhypusine formation, probably through a different mechanism.
* A preliminary account of this work has been presented (Wolff, The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. The abbreviations used are: eIF-4D, eukaryotic initiation factor 4D; DFMO, a-difluoromethylornithine; HTC, hepatoma tissue culture; CHO, Chinese hamster ovary; dansyl, 5-dimethylaminonaphthalene-1-sulfonyl. hydroxylated (2,3). The enzyme that catalyzes the final step, deoxyhypusine hydroxylase, has been partially purified and characterized (4).
The mechanism of deoxyhypusine synthesis was studied in mammalian cells by the use of specifically isotopically labeled precursors, i.e. spermidine and lysine (2, 3, 5 ) . Based on the knowledge obtained from these studies as to the origin of the atoms of hypusine and evidence that one of the two protons on carbon 5 of spermidine is abstracted during transfer of the 4-aminobutyl moiety (6), we proposed a biosynthetic pathway that involves the oxidative cleavage of spermidine between carbon 5 and the secondary amino nitrogen (Scheme 1). The detection of free 1,3-diaminopropane, which was postulated to form in parallel with butylamine transfer, was not possible in cultured cells, however, due to the very low consumption of spermidine for hypusine synthesis (0.01-0.02%/h) (8)(9)(10) and the diverse metabolic pathways of polyamines.
Earlier attempts to develop a cell-free system for deoxyhypusine synthesis were unsuccessful, partly because the protein substrate is lacking in cells that contain a normal level of polyamines. Recent data suggest that a precursor (Mr -18,000, PI 5.1) of eIF-4D (Mr -18,000, PI 5.3) accumulates only upon depletion of spermidine (8). By treatment of cells with DFMO (ll), an irreversible inhibitor of the initial enzyme of polyamine biosynthesis, ornithine decarboxylase, spermidine can be lowered to a level that depresses deoxyhypusine synthesis and causes accumulation of the protein substrate (8). Murphey and Gerner (12) recently accomplished cell-free synthesis of deoxyhypusine in lysates of DFMOtreated HTC cells and showed that the reaction is critically dependent on pH (optimum at 9.5). Under similar reaction conditions, we observed the synthesis of deoxyhypusine in lysates of DFMO-treated CHO cells. We report here a separation of the enzyme and the protein substrate by ammonium sulfate fractionation, a further purification of the protein substrate, and stimulation by NADf and certain other pyridine nucleotides. Using the reconstituted system, from which most of the cellular amines and other low molecular weight compounds had been removed, we were able to determine the fate of the propylamine moiety of spermidine during deoxyhypusine synthesis and have examined the effects of spermidine analogues and other amines on this reaction. Spande (National Institutes of Health). [terminal meth~knes-~H] Spermidine' was purchased from Du Pont-New England Nuclear; 1,3-diaminopropane, 1,5-diaminopentane, 1,6-diaminohexane, 1,7diaminoheptane, l,S-diaminooctane, l,9-diaminononane, 1,lO-diaminodecane, N'-benzylspermidine, aminoguanidine, benzoyl chlobis(guanylhydrazone), NAD, NADH, and NADPH from Sigma; ride, isoniazid, quinacrine, apdipyridyl, ATP, methylglyoxal NADP from Calbiochem; pyrroloquinoline quinone from Fluka; dansyl chloride from Pierce Chemical Co.

Methods
Fractionation of the Protein Substrate and the Enzyme(s)-CHO cells were cultured as outlined previously (3). Additions of DFMO (4 mM) were made to subconfluent rapidly growing cells at a density of approximately 1-2 X lo6 cells/100-mm dish, and cells were grown in the presence or absence of DFMO for 42 h. After the incubation, cellular polyamine levels were measured. Since the accumulation of the protein substrate critically depends on spermidine deficiency, the batches of cells virtually depleted of spermidine (C0.2 nmol/mg protein) were used as DFMO-treated samples. Subsequent operations were carried out at 4 "C. Untreated or DFMO-treated cells were washed twice with phosphate-buffered saline (pH 7.4) and harvested by centrifugation. Lysates were prepared using cells from 100-600 dishes by suspending and sonicating (60 s at 70 watts) cells in 10-60 ml (0.1 ml of buffer per 1 X lo7 cells) of phosphate-buffered saline.
The cellular debris were removed by centrifugation for 30 min at 25,000 X g. The clarified lysate was treated with ammonium sulfate (0.277 g/mi) to achieve 45% saturation, and the precipitates were collected by centrifugation. The resulting supernatant solution was treated with solid ammonium sulfate (0.21 g/ml) to achieve the final concentration of 75% saturation. The precipitates from the 0-45% [terminal methykne~-~H]Spermidine is [1,8-3H]spermidine according to the numbering system first used by Tabor et al. (7). See Scheme 1. and 45-75% ammonium sulfate fractions were dissolved in 2-12 ml of phosphate-buffered saline (0.02 ml per 1 X lo7 cells used for the original lysate preparation) and dialyzed against 2 liters of the same buffer for 3 h with two changes of buffer. The dialyzed fractions were frozen in aliquots and stored at -20 "C. Further purification of the protein substrate was carried out by chromatography on a DEAE-Sephacel column as described in the legend to Fig. 2.

p~) ,
10-25 p1 of the 0-45% (NH4)2SO4 fraction from untreated or DFMO-treated cells as enzyme, and 10-25 pl of the 45-7576 (NH4)'S04 fraction from DFMO-treated cells or purified column fractions (Fig. 2) as protein substrate in a final volume of 70-150 pl. incubations were conducted at 37 ' C for 2 h. The reactions were terminated by the additions of an equal volume of 20% trichloroacetic acid. When the purified substrate was used, 500 pg of bovine serum albumin was added as carrier protein. The supernatants were separated, and the precipitates were washed three times with 5% trichloroacetic acid containing 1 mM each putrescine, spermidine, and spermine in order to remove noncovalently bound labeled polyamines. The washed precipitates were hydrolyzed in 6 N HCl for 18 h at 106 "C, and radioactivity in deoxyhypusine was measured after separation of this amino acid by ion exchange chromatography as described previously (13, 14). Specific details of each experiment are given in the figure and table legends.

Cell-free Synthesis of Deoxyhypusine: Separation of Protein Substrate and Enzyme Fractions--Rapid synthesis of [3H]
deoxyhypusine was observed when crude lysate prepared from DFMO-treated cells, postmitochondrial supernatant from this lysate (25,000 X g, 30 min), or a dialyzed 0-75% ammonium sulfate fraction of the lysate was incubated with [1,8-3H] spermidine in pH 9.5 buffer. Hypusine was not formed because the enzyme responsible for conversion of deoxyhypusine to hypusine is inactive at this high pH (4,12). In each case deoxyhypusine synthesis occurred in an -18,000-dalton protein (data not shown). Unlike the lysates from DFMO-treated cells, lysates from untreated cells did not support deoxyhypusine synthesis.
The data in Fig. 1 illustrate a clean, simple separation of the protein substrate and the enzyme(s) and further show that the failure of the lysate of untreated cells to synthesize deoxyhypusine is due to the lack of the protein substrate. Little, if any, deoxyhypusine synthesis was observed in a reaction mixture containing a single fraction from either untreated or DFMO-treated cells ( Fig. 1, A , B , D, and E ) , in the same ammonium sulfate fractions from both cells after their combination (G and J ) , or in any combination of two fractions that did not include the 45-75% ammonium sulfate fraction from DFMO-treated cells (F and I). Only two combinations gave efficient deoxyhypusine synthesis: the 45-75% ammonium sulfate fraction from DFMO-treated cells with the 0-45% ammonium sulfate fraction from untreated cells ( H ) or from DFMO-treated cells (C). In view of previous evidence for the presence of an eIF-4D precursor in DFMOtreated cells but not in untreated cells (8), it follows that the 45-75% ammonium sulfate fraction from DFMO-treated cells contains the critical protein substrate. This is consistent with the fact that mature eIF-4D with approximately the same molecular weight as the precursor is precipitated between 45 and 75% ammonium sulfate. On the other hand, it seems that the 0-45% ammonium sulfate fractions of both untreated and DFMO-treated cells contain the enzymatic activity for deoxyhypusine synthesis.
Partial   (17). Although we had observed a 3-4fold stimulation of deoxyhypusine synthesis by 1 mM NAD in the combined ammonium sulfate factions, partial purification of the substrate protein from the 45-75% ammonium sulfate fraction led to an almost absolute requirement for NAD+ in this reaction. The effects of increasing concentrations of NAD' and the comparison of various cofactors added at 1 mM are shown in Fig. 3. 1,3-Diaminopropane as a Product of the Deoxyhypusinesynthesizing Enzyme(s)-When the partially purified protein substrate was used for deoxyhypusine synthesis, no interconversion of polyamines was observed. Labeled putrescine was not utilized for deoxyhypusine formation using this substrate preparation (data not shown), confirming that spermidine is the direct polyamine precursor of deoxyhypusine (2,5,6).

Fraction No
FIG. 2. Partial purification of the protein substrate on DEAE-Sephacel. To 6.5 ml of dialyzed 45-75% ammonium sulfate fraction from DFMO-treated cells, containing 89,000 cpm of 13H] hypusine-labeled eIF-4D as tracer, 8.5 ml of 0.05 M Tris acetate buffer, pH 6.8, 1 m M dithiothreitol, 0.1 mM EDTA was added. It was applied to a 6-ml column of DEAE-Sephacel (40-150-pm particle size) that had been equilibrated with the buffer. After washing with 12 ml of buffer, a gradient from 0 to 0.5 M KC1 in the same buffer was applied, and fractions of 1.5 ml were collected. Fractions containing the substrate protein (29-40) were pooled after determining the substrate activity in each fraction. Each reaction mixture contained 5 pCi of [3H]spermidine, 10 p1 of 0-45% ammonium sulfate fraction from DFMO-treated cells as enzyme, 30 p1 of substrate fraction, and 0.5 mM NAD+ in a final volume of 100 pl. Other conditions were as described under "Experimental Procedures." Only the butylamine portion of spermidine is utilized as a structural component for deoxyhypusine synthesis (2,3). The data presented in Fig. 4 and Table I provide strong evidence that transfer of the butylamine moiety t o a specific lysine residue to form deoxyhypusine involves the cleavage of spermidine to yield free 1,3-diaminopropane (Scheme 1). Under the conditions given in Fig. 4 for the cell-free synthesis of deoxyhypusine, -12% of the total spermidine was converted to deoxyhypusine, thus permitting detection of the other cleavage product. Fig. 4A (0 (0) 10 pl of 0-45% ammonium sulfate fraction from DFMO-treated CHO cells as enzyme, 5 pCi of [1,8-3H]spermidine (1.53 p~) , 0.5 mM NAD+ in a final volume of 100 pl. After incubation, the protein was precipitated with trichloroacetic acid as described under "Experimental Procedures." Ion exchange chromatography was carried out as described earlier (13, 14) using elution with sodium citrate buffer (1.5 N Na+, pH 5.55) for 10 min and then with sodium citrate buffer (3.0 N Na+, p H 5.55) for 30 min, followed by re-equilibration with the initial buffer for 17 min. One-minute fractions were collected and the radioactivity measured. The counts/ min shown correspond to -5% of the total sample. The elution positions of deoxyhypusine, 1,3-diaminopropane, and the polyamines are indicated by arrows. A , acid hydrolysate of trichloroacetic acidprecipitated protein. The observed radioactivity in the protein hydrolysate was corrected for losses of [3H]deoxyhypusine-containing protein during precipitation, washing, and transfer to permit direct comparison with part B. B, trichloroacetic acid supernatant. Radioactivity in fractions 1-20 was corrected for impurities present in commercial [3H]spermidine. The radioactivities in spermidine were 229,000 and 232,000 cpm, respectively, in the reaction mixture containing purified substrate with (0) and without (0) enzyme.
This component rechromatographed at its original position following treatment with acid (6 N HC1, 18 h at 106 "C), thus ruling out the possible identity or contamination with N1and NE-acetylspermidines or other spermidine derivatives. The amount of radioactivity in this component was approximately the same as that in the protein-bound deoxyhypusine. Various compounds, e.g. guazatine and N4-benzylspermidine, that inhibited deoxyhypusine synthesis (Fig. 5) caused a parallel suppression in the production of this labeled component (data not shown). Neither this labeled component at the position of 1,3-diaminopropane nor [3H]de~xyhypu~ine was formed in the absence of enzyme (Fig. 4, A and B, 0). Table I summarizes the results of thin-layer chromatography and high performance liquid chromatography of the dansyl and benzoyl derivatives, respectively, of 1,3-diaminopropane, putrescine, and the radioactive component that forms in parallel with deoxyhypusine. The fact that the derivatives of 1,3-diaminopropane and the radioactive compound chromatograph identically is strong evidence for this compound as 1,3-diaminopropane.

Identification of 1,3-diaminopropane by chromatography of its dansyl
and benzoyl derivatives Cell-free synthesis of deoxyhypusine and chromatography of the trichloroacetic acid supernatant were carried out as described in the legend to Fig. 4. The fraction of labeled material that eluted from the column at the position of 1,3-diaminopropane (fraction 15, Fig. 4B, 0 ) was mixed with 50 nmol of unlabeled 1,3-diaminopropane. A portion of this mixture was treated with dansyl chloride, and the derivative was chromatographed on silica gel 60 (18). The dansyl derivatives were localized under uv light. The location of radioactivity was determined by removing 2-3-mm segments of the layer along the development track and examining each for radioactivity in liquid scintillation fluid. The benzoyl derivative was prepared using another portion of the mixture, and separation was conducted by high performance liquid chromatography according to Redmond and Tseng (19) using pBondapak CI8 column (Waters) and isocratic conditions (methanol:water, 58:42) and a flow rate of 1.5 ml/min. Absorbance was monitored a t 254 nm, and 0.5-min fractions were collected for measurement of radioactivity. The abbreviations used are: DAP, 1,3diaminopropane; PTC, putrescine; DAH, 1,6-diaminohexane; SPD, spermidine; TCA, trichloroacetic acid.
Several inhibitors of amine oxidases were found to be less effective inhibitors than 1,3-diaminopropane or the spermidine analogues tested. Guazatine, an inhibitor of a plant amine oxidase (21) that catalyzes a similar spermidine cleavage, was reported by Murphey and Gerner (12)  HTC cells. We found similar degrees of inhibition by guazatine and by pyrroloquinoline quinone, the cofactor of serum spermine oxidase in our cell-free system (-30-40% at lo-* M, >90% at lov3 M). Methylglyoxal bis(guany1hydrazone) and aminoguanidine caused approximately 50% inhibition at M. Quinacrine, isoniazid, and the metal chelators &,a-dipyridyl and 1,lO-phenanthroline showed no inhibition.

DISCUSSION
The data reported here represent a significant step toward the purification and characterization of the enzyme(s) responsible for deoxyhypusine synthesis and of its protein substrate, and provide further insight into the mechanism of this posttranslational modification reaction. We have accomplished the initial separation of the enzyme and the protein substrate, further purified the substrate, demonstrated a cofactor requirement for deoxyhypusine synthesis, and identified 1,3diaminopropane as a spermidine cleavage product formed in conjunction with the butylamine transfer.
The use of an in vitro system for deoxyhypusine synthesis has permitted us to separate and partially purify the protein substrate. This protein is likely to be the eIF-4D precursor (M, 18,000, PI 5.1) recently identified on the two-dimensional map of proteins of DFMO-treated CHO cells (20) as evidenced by its similar ion exchange chromatographic properties to eIF-4D, shown in Fig. 2. The results presented here are also consistent with mevious reDorts from this laboratory. which in DFMO-treated CHO cells (8), and further support the notion that accumulation of the eIF-4D precursor that serves as an acceptor of the 4-aminobutyl moiety from spermidine occurs upon spermidine depletion (20).
Because [1,8-3H]spermidine, in which the tritium is distributed equally between the two terminal methylenes, was used for deoxyhypusine synthesis, equal amounts of radioactivity would be expected in the two products. That approximately the same amount of radioactivity was found in 1,3-diaminopropane and deoxyhypusine is in accordance with the proposed cleavage between C5 and the secondary amino nitrogen of spermidine as the initial step in hypusine biosynthesis (Scheme 1). A cleavage of spermidine at the same position to produce 1,3-diaminopropane and 4-aminobutyraldehyde, which spontaneously cyclizes to A'-pyrroline, has been described in oat seedlings (21) and Serratia marcescens (22), but has not yet been detected in animals. Since little radioactivity is found in the early fractions (Fig. 4B) where A'-pyrroline is expected to elute, it is likely that 1,3-diaminopropane is produced in a coupled reaction with deoxyhypusine formation in our cell-free system. In the mechanism in Scheme 1 the formation of a proposed Schiff base between the 4-aminobutyraldehyde, or other intermediate, and the t-amino group of a specific lysine residue of the protein substrate may occur without dissociation of the intermediate from the enzyme and with provision for the prevention of nonenzymic cyclization to A'-pyrroline. The formation of the initial intermediate, whether an aldehyde or not, probably involves removal of a proton from carbon 5 adjacent to the cleavage site in spermidine. This is evident from labeling studies in intact CHO cells that showed the loss of one of the hydrogens from carbon 5 of spermidine during its conversion of deoxyhypusine (6). In view of the stimulation by NAD+ it is tempting to postulate a role for a pyridine nucleotide at this step. The reduction of the double bond of the postulated Schiff base may be accomplished at a later stage by a different pyridine nucleotide or by another coupled cofactor.
The inhibition studies with various amines also shed light on this unique enzymatic reaction. In view of the evidence for the coupled formation of 1,3-diaminopropane and deoxyhypusine, the potent inhibition by 1,3-diaminopropane may be postulated as product inhibition. On the other hand, the inhibition by compounds that structurally resemble spermidine may be caused by competition at the substrate amine binding site. The strong inhibition by l&diaminooctane, 1,7diaminoheptane, homospermidine, N4-benzylspermidine, and caldine suggests a structural requirement for the two amino groups spaced at a distance close to that between the two primary amino groups of spermidine and also suggests that the secondary amino group is not necessary for binding to the active site of the enzyme.
Although the efficiency of deoxyhypusine synthesis in the combined ammonium sulfate fractions (-4 pmol/mg protein in 2 h) is comparable to the maximum rate observed in DFMO-treated intact CHO cells (-8 pmol/mg protein in 2 h) (a), some variations were observed in different preparations.
Loss or dissociation of NAD+ or other cofactors is probably the cause of this variability. Whether NADC is the physiological cofactor for this reaction and, if so, in which step of the reaction it participates, remains to be determined after further purification of the enzyme.